github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 13:19:40 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #4687] Phi-3-medium is degraded #2950

New Issue
Closed
opened 2026-04-12 13:19:39 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 13:19:39 -05:00
Owner
Copy Link

Originally created by @pdevine on GitHub (May 28, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4687

          Conversely, Phi-3-medium is degraded in the ollama version. it has lower benchmark scores than Llama.cpp. several people have tried Phi-3-medium on ollama and found the performance to be poor. Why is this happening? Either the inference parameter settings or the chat template is wrong or something is broken in quantization.

Originally posted by @umiyuki in https://github.com/ollama/ollama/issues/4626#issuecomment-2130726686

Originally created by @pdevine on GitHub (May 28, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4687 Conversely, Phi-3-medium is degraded in the ollama version. it has lower benchmark scores than Llama.cpp. several people have tried Phi-3-medium on ollama and found the performance to be poor. Why is this happening? Either the inference parameter settings or the chat template is wrong or something is broken in quantization. _Originally posted by @umiyuki in https://github.com/ollama/ollama/issues/4626#issuecomment-2130726686_
GiteaMirror commented 2026-04-12 13:19:41 -05:00
Author
Owner
Copy Link

@tinycrops commented on GitHub (May 29, 2024):

I'm getting this message when I try to use it

Error: llama runner process no longer running: -1
<!-- gh-comment-id:2136334979 --> @tinycrops commented on GitHub (May 29, 2024): I'm getting this message when I try to use it ``` Error: llama runner process no longer running: -1 ```
GiteaMirror commented 2026-04-12 13:19:42 -05:00
Author
Owner
Copy Link

@zane29 commented on GitHub (May 29, 2024):

For Phi3:14b,macOs can working.ollama version is 0.1.39

<!-- gh-comment-id:2136366955 --> @zane29 commented on GitHub (May 29, 2024): For Phi3:14b,macOs can working.ollama version is 0.1.39
GiteaMirror commented 2026-04-12 13:19:43 -05:00
Author
Owner
Copy Link

@yumemio commented on GitHub (May 30, 2024):

Related?: https://x.com/MatthewBerman/status/1793386069389623323

ollama @ollama
May 23
performance reasons. Our apologies. In this case, we believe this is very deep in the inference code. (ie. operation being done wrong or the kv cache).

We've been working on updating a lot of the inference side to patch up our dependencies -- it's taking us some time though.

Also some folks in llama.cpp are reporting degraded performance: https://github.com/ggerganov/llama.cpp/issues/7520

Not sure this is an upstream (llama.cpp) issue or something specific to Ollama.

<!-- gh-comment-id:2138751836 --> @yumemio commented on GitHub (May 30, 2024): Related?: https://x.com/MatthewBerman/status/1793386069389623323 > [ollama](https://x.com/ollama) [@ollama](https://x.com/ollama) > [May 23](https://x.com/ollama/status/1793389631414632642) > performance reasons. Our apologies. In this case, we believe this is very deep in the inference code. (ie. operation being done wrong or the kv cache). > > We've been working on updating a lot of the inference side to patch up our dependencies -- it's taking us some time though. Also some folks in llama.cpp are reporting degraded performance: https://github.com/ggerganov/llama.cpp/issues/7520 Not sure this is an upstream (llama.cpp) issue or something specific to Ollama.
GiteaMirror commented 2026-04-12 13:19:44 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Jul 11, 2024):

I think it's OK to close this now. It seems to be working pretty well in the most recent versions of Ollama:

I'm getting this performance on a MBP which feels inline with other models around the same size.

prompt eval count:    9 token(s)
prompt eval duration: 379.595ms
prompt eval rate:     23.71 tokens/s
eval count:           430 token(s)
eval duration:        12.433999s
eval rate:            34.58 tokens/s
<!-- gh-comment-id:2221910387 --> @pdevine commented on GitHub (Jul 11, 2024): I think it's OK to close this now. It seems to be working pretty well in the most recent versions of Ollama: I'm getting this performance on a MBP which feels inline with other models around the same size. ``` prompt eval count: 9 token(s) prompt eval duration: 379.595ms prompt eval rate: 23.71 tokens/s eval count: 430 token(s) eval duration: 12.433999s eval rate: 34.58 tokens/s ```
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#2950
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?